Re: [Isis-wg] [rbridge] Why is MTU discovery important?

[Dave Katz <dkatz at juniper.net>]

This is exactly the same problem that NLSP ran into 15 years ago  
(multiple encaps with different MTUs on Ethernet media.)  NLSP punted  
by requiring configuration of the max LSP size (IIRC), and there is  
existence proof <cough> that this is a non-starter.

--Dave


On Apr 15, 2009, at 8:21 AM, Sadler, Jonathan B. wrote:

> Exactly my point.  The "payload guaranteed" is different depending  
> on if you are looking at 802.1d or 802.1ad.  A 802.1ad "1500 byte  
> payload" will not make it through an 802.1d switch as the 802.1d  
> switch considers the 802.1Q shim as part of the payload - making the  
> overall frame size larger than the 1500 byte max it is required to  
> support.
>
> I agree there are "lax" devices out there.  Needless to say, the  
> need for "MTU discovery" comes from those that limit MTU to the 1500  
> byte definition found in 802.1d.
>
> Jonathan Sadler